Servers by jDrones

Comparing a uBlox Zed-F9P GPS with a M8P without external corrections

(tridge) #1

Google Photos
As the new uBlox Zed-F9P GPS is starting to become available I thought it would be worthwhile to do some side by side testing with the previous generation uBlox M8P. This testing is without external corrections, so both GPS modules are running as standalone GPS units.
The key change with the F9P is that it is dual-band when paired with a dual-band antenna. That allows is to estimate ionospheric corrections much more accurately than can be done with a single band GPS module like the M8P.

Test Setup

GPS modules used were:

The two modules were connected to the same Hex CubeBlack running ArduPilot, with the M8P GPS as GPS1 and the F9P as GPS2.
Note that the external antenna, long antenna cable and magnetic base of the F9P setup make it impractical for most small aircraft, but modules with integrated multi-band patch antennas are now starting to become available. The first of those is here:

I don’t have one of those yet (it wasn’t available when I ordered the Drotek unit), so I tested with the one I had. I look forward to other vendors releasing more F9P based GPS modules.

Scenarios

Three scenarios were tested

  • a extremely poor location looking out the window of a house with a metal roof, looking over another house. Clear view of the sky is only available over about 20 degrees vertically. This represents extreme multi-pathing.
  • a good location with clear sky view
  • a medium location with medium density tree cover and a nearby house

All 3 locations were in Canberra, Australia.

Even though these GPS modules are capable of RTK fixes a source of corrections was not supplied. The aim of these tests was to see if the dual band and improved processing in the F9P would produce less variance in position and altitude without external corrections.

Location 1:

The first location was with extreme multi-pathing. The antennas were up against a window in a house with a metal roof and metal eaves, looking over another house.

First we show the altitude reading in meters over a 12 hour period:

And now the horizontal error in meters:

Location 2:

This location was at the CMAC flying field, with clear view of the sky in all directions. Altitude in meters over 3 hours:

and horizontal error in meters:

Location 3

This location was in the backyard of my house, with medium density tree cover. Altitude error in meters over 3 hours:

and horizontal error in meters:

Conclusion

The F9P with a dual-band antenna provides a very large reduction in both altitude drift and horizontal drift over a single band M8P. Altitude drift is still high enough with the F9P to make using a rangefinder or other altitude source for landing worthwhile, but mission accuracy should be a lot better.
The F9P reported approximately twice as many satellites as the M8P in each situation, with up to 30 satellites tracked with a clear view of the sky. It seems likely that it was the same satellites as the M8P with each reported once for each band (L1 and L2).

14 Likes
Comparing a uBlox F9P with a M8P with RTK
Dev Call June 10 2019 2300 UTC
Comparing a uBlox F9P with a M8P with RTK
(rmackay9) #2

Well, that’s quite an improvement! Great to have a real back-to-back comparison.

1 Like
(Mike Boland) #3

Will be interesting to see how it pans out with RTK corrections.

1 Like
(Kenny Trussell) #4

Thanks for doing this. Very impressive. I have 2 F9P eval boards from Ublox but have not had time to do anything with them yet. I have been very satisfied with my C94-M8P setup but now you have me wanting to see the difference!

1 Like
(MartinKeilloh) #5

do you have a link were to purchase these gps’s please

(Tegwin) #6

We are using the F9P with RTK corrections and can say they are fantastic. Millions of times better than the M8P!

We are currently using the Drotek f9P module here with a generic multiband antenna. here

1 Like
(tridge) #7

I’ve done a followup article on RTK performance of the two modules:

2 Likes
(Robo Roby) #8

Nice work @tridge – good to see these modules taking off.

1 Like
(Jeff) #9

Very informative, had no idea that variance would be so high - thanks for sharing!

1 Like
(Michael Nunan) #11

I know a source for dual (and triple!) band antenna. Tallysman Wireless - I have used in the past - the have a wide range of antennas

see https://www.tallysman.com/index.php/gnss/products/

(Rolf) #12

Hi Tridge,
Did you measure with EGNOS enabled on M8P ? (If egnos datas are available).

(Tore) #13

Hi, nice test. So both antennas was used as is, without any groundplane?

(tridge) #14

no EGNOS in Australia

(tridge) #15

right. I’m planning to do more testing in the future with other antenna setups

1 Like
(Liang Tang) #16